Exploration and Assessment of Event Data

Abstract

Event data is generated in many domains, like business process management, industry or healthcare. These datasets are often unstructured, exhibit variant behaviour, and may contain errors. Before applying automated analysis methods, such as process mining algorithms, the analyst needs to understand the dependency between events in order to decide which analysis method might fit the recorded events. We define a categorization scheme of event dependencies and describe a preliminary approach for exploring event data, combining visual exploration with pattern mining. Events of interest can be selected, grouped, and visually explored, using either a sequential or a temporal scale. We present two use cases with shopping event data and report expert feedback on our approach.